Initial commit

grep.c

+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <string.h>
+#include <sys/stat.h>
+
+#include "regex.h"
+
+const char *ANSI_RESET_COLOR = "\033[39m\033[49m";
+const char *ANSI_COLOR_PATTERN = "\033[38;2;%d;%d;%dm";
+
+void search_file(char *regex, char *filename) {
+    FILE* f = fopen(filename, "r");
+    if (!f) {
+        fprintf(stderr, "File %s does not exist!\n", filename);
+        exit(2);
+    }
+
+    struct stat fstats;
+
+    int result = fstat(fileno(f), &fstats);
+    if (result) {
+        fprintf(stderr, "Could not find the size of the file %s\n", filename);
+        exit(2);
+    }
+
+    size_t file_length = fstats.st_size;
+
+    char *contents = malloc((file_length + 1) * sizeof(char));
+    fread(contents, file_length, sizeof(char), f);
+    contents[file_length] = '\0';
+
+     char *brkt;
+     for (char *line = strtok_r(contents, "\n", &brkt); line; line = strtok_r(NULL, "\n", &brkt)) {
+         if (!line) {
+             puts("\n");
+         }
+         else {
+             range_t *result = regexp_match(regex, line);
+             if (result) {
+                 char *to_print = calloc(strlen(line) + 1 + (strlen(ANSI_COLOR_PATTERN) + 3) + strlen(ANSI_RESET_COLOR), sizeof(char));
+                 strncpy(to_print, line, result->start);
+                 snprintf(to_print + result->start, strlen(ANSI_COLOR_PATTERN) + 3, ANSI_COLOR_PATTERN, 0, 128, 0);
+                 strncat(to_print, line + result->start, result->end - result->start);
+                 strcat(to_print, ANSI_RESET_COLOR);
+                 strcat(to_print, line + result->end);
+                 printf("%s\n", to_print);
+             }
+             free(result);
+         }
+     }
+     free(contents);
+     fclose(f);
+}
+
+int main(int argc, char *argv[]) {
+    if (argc < 3) {
+        fprintf(stderr, "USAGE: %s <regexp> filepaths...\n", argv[0]);
+        exit(1);
+    }
+
+    char *regex = argv[1];
+    for (size_t i = 2; i < argc; i++) {
+        search_file(regex, argv[i]);
+    }
+}

regex.c

+#include <stdlib.h>
+#include <string.h>
+#include <stdbool.h>
+#include <assert.h>
+#include <stdio.h>
+
+#include "regex.h"
+
+/*
+ To implement:
+ - ^
+ - $
+ - [...]
+ - [^...]
+ - *
+ */
+
+range_t*_regexp_match(char *whole_regex, char *regex, char *line, bool caret, bool dollar, size_t index, size_t start, size_t len) {
+    if ((regex[0] == '\0' && line[0] == '\0') || (regex[0] == '\0' && !dollar)) {
+        range_t *result = malloc(sizeof(range_t));
+        *result = (range_t){.start=start, .end=len + start};
+        return result;
+    }
+    else if (regex[0] == '\0') {
+        return NULL;
+    }
+    else if (regex[1] == '*') {
+        if (regex[0] == line[0] && line[0] != '\0') {
+            range_t *end = _regexp_match(whole_regex, regex + 2, line + 1, caret, dollar, index + 1, start, len + 1);
+            if (end) {
+                return end;
+            }
+            range_t *include = _regexp_match(whole_regex, regex, line + 1, caret, dollar, index + 1, start, len + 1);
+            if (include) {
+                return include;
+            }
+        } 
+        if (!caret && line[0] != '\0') {
+            range_t *restart = _regexp_match(whole_regex, whole_regex, line + 1, caret, dollar, index + 1, index + 1, 0);
+            if (restart) {
+                return restart;
+            }
+        }
+        range_t *skip = _regexp_match(whole_regex, regex + 2, line, caret, dollar, index, start, len);
+        if (skip) {
+            return skip;
+        }
+        return NULL;
+    }
+    else if (regex[0] == line[0]) {
+        return _regexp_match(whole_regex, regex + 1, line + 1, caret, dollar, index + 1, start, len + 1);
+    }
+    else if (!caret && line[0] != '\0') {
+        return _regexp_match(whole_regex, whole_regex, line + 1, caret, dollar, index + 1, index + 1, 0);
+    }
+    else {
+        return NULL;
+    }
+}
+
+range_t *regexp_match(char *regex, char *line) {
+    bool caret = false;
+    if (regex[0] == '^') { 
+        caret = true; 
+        regex += 1; 
+    }
+
+    bool dollar = false;
+    if (regex[strlen(regex) - 1] == '$') { 
+        dollar = true; 
+        regex[strlen(regex) - 1] = '\0';
+    }
+    assert(line[0] != '*');
+    return _regexp_match(regex, regex, line, caret, dollar, 0, 0, 0);
+}

regex.h

+#ifndef __REGEX_H__
+#define __REGEX_H__
+
+#include <stdint.h>
+
+typedef struct range {
+    size_t start;
+    size_t end;
+} range_t;
+
+range_t *regexp_match(char *regex, char *line);
+
+#endif // #ifndef __REGEX_H__